TalaskwelaPhilippines
Built by a Filipino parent

A child’s school sends everything, everywhere.

We tell parents only what matters, what they have to do, and what their child needs next.

The parent screenshots the school dashboard, or photographs the page in their child’s notebook, and uploads it. Nothing is saved until they approve it.

Not deployed. No real family data has been entered.

1Upload
2Read
3You approve
Screenshot from the school dashboard
DARB · Thursday, 3 September
LanguageDrill Sheet #1: Homophones, tomorrow, 4 September.ScienceDo Datafile #2, due 10 September.MathReview multiplication drills.FilipinoBasahin ang pahina 28.MusicSee the class feed.
We pulled this out

4 things to do. 1 flagged for you to check.

1 of 3 · School dashboard
You’re in control
The problem

School information arrives unsorted.

Most of it is posted to the school dashboard. Some of it is a form to fill in, or a PDF to print, sign, scan and send back. Some exists only as a line your child copied into a notebook. Nothing tells a parent which ones carry an obligation, which ones are context, and which ones say something about how their child is doing.

One upload, sorted by what it is.
exam advisory
form to sign
lesson handout
reading list
graded quiz
graded seatwork
2 things to do - exam advisory, form to sign
2 stored as context
2 kept for mistake analysis
How it works

What happens after a parent uploads a photo.

1

Upload it

A screenshot of the school dashboard, a photo of the notebook page, or a form the school wants signed.

2

The engine reads it

It identifies what the document is, then pulls out the subject, what is due, and when.

3

Approval, then a reminder

Nothing is stored as a task until the parent taps Approve. Once approved, it carries a reminder.

Three routes, one upload

Most uploads never become a task.

The product first works out what kind of document it is. Each kind goes somewhere different.

01

Notices, advisories, announcements

Turned into things the parent must do, with a date, a reminder, and a link back to the photo it came from.

Becomes a to-do
02

Lesson notes and study material

Stored as context about what the child is currently being taught. Never turned into tasks.

Stored as context
03

Graded papers

Kept so that mistakes can be analyzed later, once the results layer exists. Never turned into tasks.

Kept for later analysis

Two thirds of what a parent uploads never becomes a to-do.

The model

How the company makes money.

14-day free trial

DONE

Can I try it first?

Every feature is unlocked for 14 days, with no usage limit and nothing held back. It is a trial, not a free plan - there is no permanent free tier.

Inviting another parent extends the trial by 14 days, up to 42 days in total.PLANNED

Paid family plan

IN BUILD

What does my child need?

One child’s own history, their marks, and the concepts they keep missing, and what to practice next. The organizer, reviewer and quiz are built. The practice that adapts to a child’s own record is in development.

₱399 a month for one child, or ₱3,990 a year. Every paid plan carries every feature - the price depends on how many children you have, never on which features you get.
Later

These are not projections. They show where the model could go later.

Shared class intelligencePLANNED
What is happening?

One notice read once could serve every parent in a class, which is what would make a free layer cheap enough to offer at all. It is Phase 4, and it is planned rather than decided.

AgenticPLANNED
Handle it for me.

It would act on the parent’s behalf rather than telling them what to do. It is not built and not scoped for the near term.

Two children at ₱549 and three or more at ₱699 are proposed, not settled.

Defensibility

What makes this hard to copy.

Anyone can read a notice. What accumulates is one child’s record, built up over terms and years.

A practice set built from one child’s wrong answers has no value to another child. That history cannot be copied and cannot be bought.

Reading a notice
It is a commodity that anyone can do.
One child’s marks over time
It accumulates only here.
The mistakes they repeat
It accumulates only here.
Practice built from those mistakes
It cannot be transferred anywhere else.
Trust and control

Six commitments built into the product.

Approval is required

Nothing is saved until the parent taps Approve.

Every item shows its source

It points back to the photo it came from.

Uncertainty is flagged

Unclear documents are flagged rather than guessed at.

No child accounts

There is nothing for a child to log into. The parent holds the phone.

Consent per child, per purpose

Asked before anything is used, and can be withdrawn.

Metadata stripped before upload

Hidden location and device data are removed from photos in the parent’s own browser, before anything is uploaded.

Background

How the product developed

March 2026

First version

A Viber bot that read school messages for a single household.

July 2026

Scope review

Redefined from an assignment tracker to a full record of a child’s schooling.

August 2026

Rebuild decision

The original stack could not support consent, privacy and per-child data. It was retired.

September 2026

New foundation

Consent, permissions and data isolation built and tested before any parent-facing feature.

Today

Phase 1, saving left to do

We built the privacy and consent layer first. It cannot be added to a child’s data after the fact.

Roadmap

Where each phase is now.

Phase 0DONE

Prove extraction works, measure cost per document

Phase 1IN BUILD

The Organizer, one household

Secure database foundationDONE
Parent sign-up and onboardingDONE
Consent, permissions, withdrawalDONE
Login by email linkDONE
Upload a photo and read itDONE
Save it to the parent’s accountIN BUILD
The lists a parent seesDONE
Audit trail, every line traceable to its sourceDONE
Phase 2IN BUILD

The Reviewer and practice questions - working in the demo

Phase 3PLANNED

Results, history, adaptive practice

Phase 4PLANNED

Distribution, the free class layer

Phase 5PLANNED

Intelligence, the end goal

Phase 6PLANNED

Lifecycle extend

Where we are today

Current build status.

The engine that reads school documents works, and its cost per document has been measured.

The private foundation is built and tested: accounts, consent, permissions, withdrawal and email login. We have tests that check one family cannot see another family’s records.

Photos have their hidden metadata stripped before upload.

Nothing is saved between visits. There are no accounts yet, so the screens a parent would open the next day do not exist.

Not deployed. No real family data has been entered.